Mime-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: 8bit
Message-ID:  <l03130301b40fcf6639e9@[130.239.20.144]>
Reply-To: Mailing list for the LaTeX3 project
              <LATEX-L@URZ.UNI-HEIDELBERG.DE>
Date:         Thu, 23 Sep 1999 17:14:07 +0200
From: Lars =?iso-8859-1?Q?Hellstr=F6m?= <Lars.Hellstrom@MATH.UMU.SE>
Sender: Mailing list for the LaTeX3 project
              <LATEX-L@URZ.UNI-HEIDELBERG.DE>
To: Multiple recipients of list LATEX-L
              <LATEX-L@URZ.UNI-HEIDELBERG.DE>
Subject:      Left italic correction and fontdimens
Status: R

A while back, on the fontinst mailing list, it was noticed that there are
fonts (the concrete case was some italic fonts on the Bitstream 500 CD)
which seem to have a built-in optical justification of the margin, i.e., if
you were to unslant such a font while keeping the reference points of
character boxes fixed then the left sidebrearings would become visibly
smaller than in the corresponding upright font. The advantage with this
would be to make the margins of italic paragraphs align optically with the
margins of upright paragraphs; with most fonts they don't align. (Exercise
14.16 in The TeXbook is about achieving the same effect by fiddling with
\leftskip and \rightskip, but having it built into the fonts makes life
much simpler for LaTeX.)

The problem with such fonts come when you change to it from upright inside
a paragraph, e.g.

  ... either \emph{horizontally} or ...

or even

  ... the superthings (\emph{maximal} gadgets with ...

There should be some kern inserted---an italic correction to the left of
the italic material. As TeX only has italic corrections to the right of
characters, that mechanism will not suffice here. One thing which could be
done is to have a fontdimen which is interpreted as an italic correction to
the left, common to all characters in the font---then high-level font
changing commands (like \textit) could insert a kern of this size, just
like they currently make italic corrections to the right.

But clearly, this is a suggestion for some (probably) distant future. I do
not expect to see any implementations soon.

BTW, thinking of the above also made me think about fontdimens in general.
I noticed in the slides from TUG99 that LaTeX2e* will have symbolic names
for at least some of these, and there is clearly a need for this. Another
thing needed is the ability to test whether a font has a given fontdimen.
It would be neat if  one in FD files could set, for a font family and for
fonts individually, something which would act as "the set of fontdimens
available for this font", perhaps like one today can set the hyphenchar of
the font. This would be friendly to extensions in this area.

Finally, I'm eagerly avaiting template.dtx (or whatever you will call it).
I think templates might provide an elegant solution to some problems with
font selection in LaTeX2e, but it's hard to tell just judging from the
slides.

Lars Hellström